Infrared thermal imagers, as non-contact temperature measurement tools, have significant advantages in hot stamping processes, but their measurement accuracy is easily affected by multiple factors such as surface emissivity, observation angle, and target temperature. A temperature measurement optimization method with dynamic emissivity compensation is proposed in the article: firstly, projection measurement technology is used to accurately obtain the spatial angle parameters of complex curved parts, and the effect of observation angle and temperature values on temperature measurement deviation is quantitatively analyzed through experiments; Constructing a nonlinear mapping model between emissivity and multidimensional variables through machine learning algorithms to achieve intelligent compensation of dynamic emissivity parameters. Experimental verification shows that after compensation, the temperature measurement system error can be stably controlled within the range of ± 1.5 ℃, with an accuracy improvement of up to 60% compared to the fixed emissivity mode. This provides an effective solution for the application of high-precision infrared temperature measurement in intelligent manufacturing scenarios.